Abstract:
This paper explores two methods to reduce the computational requirements of radiative heat transfer Monte Carlo simulation. First, it presents an efficient algorithm for tracing particles in large, complex, planar geometries with nonparticipating media, detailing an efficient intersection algorithm and the implementation of Uniform Spatial Division (USD) with demonstrated efficiency through timing results. Second, it investigates improving the accuracy of Monte Carlo results by applying reciprocity and closure, using statistical theory for a reciprocity estimation smoothing (RES) technique that combines reciprocity and closure enforcement via least-squares smoothing. The paper compares several RES methods through numerous runs on two large geometries to identify the optimal approach, also examining the RES method's effects on surfaces and individual results, and deriving estimates for improvements quantifiable from a single run.
